2017年10月_cv.exp

11月 10月 09月 08月 07月 06月 05月 04月 03月 02月 01月

原创 Faster RCNN训练出现问题：smooth_L1_loss_layer.cpp:28] Check failed: bottom[0]->channels() == bottom[1]->cha

I1030 20:49:01.633868 24921 net.cpp:157] Top shape: 1 4096 (4096)I1030 20:49:01.633885 24921 net.cpp:165] Memory required for data: 280497656I1030 20:49:01.633891 24921 layer_factory.hpp:77] Creatin

2017-10-30 21:31:13 2508 3

原创 Faster RCNN训练出现问题：Selective search data not found at: /home/py-faster-rcnn/data/selective_search_dat

~/py-faster-rcnn$ ./experiments/scripts/fast_rcnn.sh 0 VGG16 pascal_voc+ set -e+ export PYTHONUNBUFFERED=True+ PYTHONUNBUFFERED=True+ GPU_ID=0+ NET=VGG16+ NET_lc=vgg16+ DATASET=pascal_voc+ arr

2017-10-30 12:54:54 2653 6

原创 caffe添加fasterrcnn_vs2013

caffe添加fasterrcnn_vs2013

2017-10-23 13:30:07 505

原创 Faster_rcnn::loadnet

bool Faster_rcnn::loadnet(){ //load net rpn_net.reset(new Net("faster_rcnn_VOC0712_ZF\\proposal_test.prototxt", caffe::TEST)); rpn_net->CopyTrainedLayersFrom("faster_rcnn_VOC0712_ZF\\proposal_fina

2017-10-21 19:27:20 416

原创 Caffe::SetDevice 设置GPU的ID号

设置GPU的ID号bool Faster_rcnn::init(){ //init //Caffe::SetDevice(0); Caffe::set_mode(Caffe::CPU); loadnet(); return true;}转载请注明：http://blog.csdn.net/forest_world

2017-10-21 19:25:01 4054

原创 Faster RCNN win版本

#include "stdafx.h"#define CPU_ONLY 1#include "Faster_rcnn.h"#include cv::Scalar colortable[20] = { cv::Scalar(0, 0, 0), cv::Scalar(0, 0, 125),cv::Scalar(0, 125, 125), cv::Scalar(125, 125, 125)

2017-10-21 18:02:27 425

原创 Faster RCNN训练成功

I1020 15:18:13.768215 2304 net.cpp:228] rpn_cls_score_reshape does not need backward computation.I1020 15:18:13.768218 2304 net.cpp:228] rpn_bbox_pred does not need backward computation.I1020 15:1

2017-10-20 15:49:52 1128

原创 Faster RCNN训练出现问题：ValueError: operands could not be broadcast together with shapes

ss = 0.0154733 (* 1 = 0.0154733 loss)I1017 02:21:18.817603 16224 solver.cpp:245] Train net output #3: rpn_loss_bbox = 0.101269 (* 1 = 0.101269 loss)I1017 02:21:18.817608 16224 sgd_solver.cpp:106

2017-10-17 03:23:57 5286

原创 Faster RCNN训练的两种模型配置

./experiments/scripts/faster_rcnn_end2end.sh 0 zf pascal_voc./experiments/scripts/faster_rcnn_end2end.sh 0 VGG_CNN_M_1024 pascal_voc

2017-10-17 02:16:41 880

原创 Python Faster R-CNN深度学习训练数据处理二

import osimport shutilimport xml.etree.ElementTree as ETimport pdb#shutil.rmtree(os.getcwd()+"/VOC2007"); os.mkdir (os.getcwd()+"/VOC2007");os.mkdir (os.getcwd()+"/VOC2007/Annotations_");

2017-10-16 21:38:17 552

原创 Python Faster R-CNN深度学习训练数据处理

import osimport shutilimport xml.etree.ElementTree as ETimport pdb#shutil.rmtree(os.getcwd()+"/VOC2007"); os.mkdir (os.getcwd()+"/VOC2007");os.mkdir (os.getcwd()+"/VOC2007/Annotations");o

2017-10-16 20:42:48 668

转载骁龙 800 系列处理器比较

骁龙 800 系列处理器比较http://www.qualcomm.cn/products/snapdragon/processors/comparison 骁龙 808处理器骁龙 810处理器骁龙 820 处理器骁龙 821 处理器处理器高达 2 GHz 的双核

2017-10-13 21:08:47 1411

原创 Faster-RCNN训练问题解决

I1013 09:20:29.058217 17752 net.cpp:270] This network produces output cls_probI1013 09:20:29.058230 17752 net.cpp:283] Network initialization done.I1013 09:20:29.122452 17752 net.cpp:816] Ignoring s

2017-10-13 12:58:58 1412

原创 python 批量修改JPG为jpg后缀

import pdbimport shutilimport osimport xml.etree.ElementTree as ET# python 3.2dir = "/home/Desktop/workpy";my_files = os.listdir(dir + "/JPEGImages_c/");os.mkdir (dir + "/JPEGImages_c_/");

2017-10-12 21:25:59 1223 2

原创 Python 文件重命名批量处理

import pdbimport shutilimport osimport xml.etree.ElementTree as ET# python 3.2dir = "/home/Desktop/workpy";def rename( files ): tree = ET.parse(dir + "/zp/" + files) root = tree.getr

2017-10-11 23:35:22 730

原创动态人脸识别技术难点

动态人脸识别技术难点光照环境人脸姿态装饰物摄像机的图像丢帧视频抖动

2017-10-11 14:44:34 3631

原创 PHP $cuserLogin->keepUser()问题 userlogin.class.php

PHP $cuserLogin->keepUser()问题 userlogin.class.php //保持用户的会话状态 //成功返回 1 ，失败返回 -1 function keepUser() { if($this->userID != '' && $this->userType != '') { global $admincachefile,$admin

2017-10-05 15:04:55 1117

原创 PHP调试方法

//连接数据库 function Open($pconnect=false) { global $dsql; //连接数据库 if($dsql && !$dsql->isClose) { $this->linkID = $dsql->linkID; } else { if(!$pconnect) { $this->linkID = @my

2017-10-04 19:58:56 441

转载图像中文描述

图像中文描述用一句话描述给定图像中的主要信息，挑战中文语境下的图像理解问题。尝试自然语言处理与计算机视觉技术结合的力量https://challenger.ai/competition/caption/subject图像中文描述问题融合了计算机视觉与自然语言处理两个方向，是用人工智能算法解决多模式、跨领域问题的典型代表。参赛者需要对给定的每一张测试图片输出一句话的描

2017-10-04 12:29:10 5418 10

转载人体骨骼关键点检测

人体骨骼关键点检测挑战复杂生活场景中的人体检测，并预测多种复杂动作下的人体骨骼关键点位置https://challenger.ai/competition/keypoint/subject简介人体骨骼关键点对于描述人体姿态，预测人体行为至关重要。因此人体骨骼关键点检测是诸多计算机视觉任务的基础，例如动作分类，异常行为检测，以及自动驾驶等等。本次比赛邀请参赛者设计

2017-10-04 12:23:02 29704 12

转载 HSL and HSV

HSL and HSVhttps://en.wikipedia.org/wiki/HSL_and_HSV#From_HSVFrom Wikipedia, the free encyclopediaContents [hide] 1Basic principle2Motivation3Formal derivation

2017-10-03 12:41:23 2742

原创 Python 图片复制

import osimport shutilimport xml.etree.ElementTree as ETdef deil_file( dirName, fileName ): #print(dirName + fileName) shutil.copyfile(dirName + fileName, "jpg/" + fileName)my_files = os

2017-10-03 03:52:08 6226

原创 Faster-RCNN训练结果

I1002 20:18:56.525288 28265 net.cpp:228] conv5_relu5_0_split does not need backward computation.I1002 20:18:56.525290 28265 net.cpp:228] relu5 does not need backward computation.I1002 20:18:56.52529

2017-10-02 20:52:07 554

原创 python生成文件图片列表

生成文件图片列表import osmy_files = os.listdir("jpg");print(my_files)listFiles = []for files in my_files: if "jpg" in files: #print(files) listFiles.append(files[:-4])listFiles.so

2017-10-02 19:21:29 2595

原创 Faster-RCNN训练问题解决：GPU内存

I1002 16:29:32.222652 27395 layer_factory.hpp:77] Creating layer bbox_predI1002 16:29:32.222658 27395 net.cpp:106] Creating Layer bbox_predI1002 16:29:32.222661 27395 net.cpp:454] bbox_pred <- fc7_d

2017-10-02 17:40:34 2665 2

原创 python实现XML文件解析与修改

import pdbimport osimport xml.etree.ElementTree as ET# python 3.2def deil_file( dirName, fileName ): tree = ET.parse(dirName + fileName) root = tree.getroot() #print(root) #fil

2017-10-02 15:41:34 2024